Vehicle Routing in complex environment

Vehicle Routing in complex environment
Coventry City Council waste collection department serves thousands of customers. Given the list of
customers (post codes) and demands/volumes of waste to be collected, it is required to design
collection routes to pick up the waste on regular bases (once or twice per week). Some of the customers
have time windows for the waste to be collected (e.g. schools have to be visited only when the pupils
are not there).
The vehicle routing problem (VRP) is an NP-hard combinatorial problem and existing algorithms allow
finding good quality solutions for only small instances.
The objective of the project is to investigate possible options of generating input data for the real life
VRPs mentioned above, and influence of the data on the robustness of the VRP algorithms. Existing
software allows to calculate distances and travel times between given points. It is obvious that time
matrices depend on the flows of traffic in the city and are subject to disruptions.
The objective of the mini-project is to suggest a way/algorithm of calculating time matrices and
investigate the influence of this data on the quality of VRP solutions.